In this era of rapid development, artificial intelligence (AI) is like an unstoppable trend, reshaping our world in its unique way. From well-known large language models such as ChatGPT, image generation diffusion models like Stable Diffusion, video generation models such as Sora, to various private intelligent models for data analysis and processing, all have rapidly evolved from emerging to empowering or even reconstructing some industries in a very short period of time. In the industrial field, numerous practical applications and future trends clearly demonstrate that AI is not only a catalyst for technological innovation but also a powerful driving force for industrial upgrading. Although for many R&D personnel, AI may still be an underdeveloped new field, even acting as a chat partner and life guide in their leisure time, the potential and capabilities of AI in product R&D have become crucial. Combining our own experience, we aim to throw some light on several application scenarios of AI in product design for readers and discuss how to bring revolutionary improvements to product design through AI. Among them, product performance indicators are usually the most important factor for a product to participate in market competition and achieve commercial success, and they are also the R&D content that leading enterprises invest the most resources in during the R&D process. Therefore, the advantages of AI in performance design are often the core scenario for the practical application of AI. In terms of product performance design, AI can predict the evolution trend of target performance indicators based on changes in input parameters. This capability helps improve R&D efficiency by reducing the number of trials and errors in the R&D process and lowering related costs. Some common application models include:
 
1. Design Scheme Output Based on Generative AI
Generative AI can automatically generate a variety of conceptual design schemes by learning historical R&D data through generative models according to the design goals, functional requirements, or preliminary sketches input by users. These schemes can include multi-dimensional design possibilities such as product appearance, structure, and materials, helping designers quickly explore different directions. For example, in automotive exterior design, by inputting key design requirements (such as aerodynamic parameters, energy consumption indicators, brand characteristics, etc.), generative AI can generate a variety of body shapes for designers to choose from and adjust.
 
2. R&D Direction Analysis Based on Public Opinion Monitoring
The ultimate goal of a product is the market, and the ultimate goal of the market is users. The concentrated feedback from users on specific products or product categories usually forms scattered public opinion information on the internet. Through AI technology to conduct public opinion analysis or sentiment analysis on data from different platforms, it can provide R&D teams with more in-depth suggestions on R&D directions. This is a key step in realizing the closed loop of the IPD methodology in the process of spiral improvement of product quality. Compared with the traditional method of collecting and analyzing problems through after-sales departments, the public opinion data summarized based on AI technology can avoid analysis errors caused by survivor bias and truly reflect the comprehensive feedback of products at the user end, which is of vital significance for product R&D and design.
 
3. Simulation Optimization and Assistance Based on AI
In the simulation and verification stage, AI can assist in simulation parameter setting and result analysis. The parameter setting in the simulation process (such as boundary conditions, physical models, and mesh division) is crucial for the accuracy of simulation results. By learning historical simulation data and expert experience, AI models can help designers optimize parameter selection and reduce the time and complexity of manual setting. Through AI-assisted analysis of simulation results (such as identifying key performance indicators and potential defects), engineers can better understand design performance and discover potential problems more quickly.
 
4. Intelligent Attenuation Model Based on AI
In the process of product R&D and design, it is necessary to derive the attenuation trend of key indicators according to stability tests to ensure that the final product can maintain specific properties or specific indicators within the validity period, and provide a basis for specific production, packaging, storage, and transportation conditions. Examples include the physical properties of drugs and specific drug components; the microbial reproduction or probiotic attenuation of food; the service life, failure rate, and reliability indicators of mechanical and electronic products; the properties and trends of material aging; and the growth and storage of crops. AI can use a large number of historical data inputs with multiple degrees of freedom to identify which input parameters will affect the attenuation trend of key indicators through deep learning technology and construct corresponding attenuation models. With the help of this AI model, in the future product design process, by adjusting input parameters, AI can generate predicted attenuation trend graphs, thereby reducing the time required for design exploration tests and improving design efficiency.
 
5. Low-Potential Design Elimination Capability Based on AI
AI data model-driven design exploration tools can not only positively guide product design and point out paths for performance improvement but also reversely eliminate design schemes with low potential. Product design based on the DOE orthogonal model is an inherent methodological tool for R&D personnel. However, through in-depth exploration and mining of large R&D data by AI, we can identify the potential impact of some non-critical indicators on the final product performance without deliberately implementing DOE design, and eliminate these non-critical factors in the design process. This enables R&D personnel to resolutely abandon low-potential designs in the early stage of the development cycle, accelerate the design iteration process, and promote product innovation.
 
6. AI-Assisted Software Performance Modeling and Prediction
AI technology, especially machine learning and deep learning, has a significant impact on software performance modeling. AI-assisted software performance modeling and prediction can become more accurate and efficient. Through step-by-step analysis, we reveal the key role of AI technology in software performance modeling and prediction. In the field of product R&D, performance, quality, and cost constitute the well-known "impossible triangle", meaning that it is difficult for these three to reach an ideal state simultaneously. R&D personnel must make trade-offs, choices, and continuous optimizations among them. However, the introduction of artificial intelligence (AI) has brought new solutions to this challenge and opened up deeper possibilities for the optimization of the three elements. This series of articles will gradually delve into the diverse application scenarios of AI in the product design process, especially how AI plays a role in product quality and cost control. Stay tuned for subsequent content.